Research into the exploration on demission clustering of site personnel of one semiconductor packaging company with two-stage clustering method

Due to the economic recession in recent years, although the total unemployed population is not retreated, the employee turnover rate in semiconductor packaging factory is frequent, and demission rate and manpower gap are high. Enterprises are always in the vicious cycle of employee demission, recruitment and education training. Under the fierce competition of globalization, semiconductor packaging industry is also under a highly competitive pressure of fast speed and short life cycle. In order to save the cost for recruitment and education training of new employees, “how to retain talents” becomes an urgent and important issue for high-tech industry to improve self-competitiveness in addition to competition in operating cost. Based on the database of 2014 new employee of case semiconductor packaging factory, this research used two-stage clustering method developed from Data Mining theory, namely Self-Organizing Maps (SOM), to find out initial solution and the best solution with K-means Method for verification comparison. The comparison between different groups was analyzed from the database to understand the differences between demission group and in-service group as the reference for the packaging factory to plan and arrange the employee recruitment in future, so as to effectively reduce the demission rate and the problem of manpower gap at bottleneck station.